‘Abracadabra’ and other AI myths

July 25, 2022 | 4 minute read
April Hawthorn
Sr. Director, Product Management, AI Apps for HCM
Text Size 100%:

In my years within HR Technology I have been a part of several vendor RFPs to select new software. When evaluating vendors for applicant tracking systems, talent management and career development, the concept of artificial intelligence has become a required “box” for any vendor to check.  What constitutes ‘artificial intelligence’, however can vary greatly from vendor to vendor.  Some offer chatbots or simple search-based matches, while other vendors provide career path recommendations or complex data analytics offerings. These same vendors give the impression that implementation and adoption of AI features is as easy as waving a magic wand. Abracadabra! The AI is implemented and your business is transformed!  

While AI is indeed an incredibly powerful tool, it is not magic.  Like any other business tool, it must run at the speed of business and provide tangible value.  With many vendors, the process of setting-up AI-powered solutions involves a prolonged period of data clean up and data creation to ensure success. Very few are integrated to other solutions (let alone the Core HR application) and require custom integrations to be built. Suffice to say, it can be hard to justify the investment and even harder to see ROI.

The real power of AI solutions for HR is how it can provide your organization value.  How will a given AI solution help you and your organization recruit more effectively or support more robust career development?  Can AI save you time in places where manual work was previously required, and can that be measured?  Will the implementation help you achieve ROI in a reasonable window of time?

When our Oracle teams began building AI solutions for recruiting, we started with a list of clear business problems that face recruiters, hiring managers and candidates. The goal was to efficiently embed pre-built AI where it could provide the most value and remove the implementation overhead to see a faster ROI. How could we “make the systems do the work” and free up recruiters and hiring managers time to focus on interviews and hiring decisions?  The answer as it turns out, is to start with some straight-forward, yet time consuming aspects of recruiting. 

  • How long will it take to hire for a particular job?: Embedded AI can provide predictions for time to hire based on historical data and learn how overall hiring times for specific roles change over time.  Understanding how long it will take to fill a role can help hiring managers better plan resources internally, pursue contracting help if needed and help forecast staff costs more accurately. Learn more about predicting time to hire
  • Reviewing high volumes of résumés for a single job.  Sifting through a large set of job applicants and resumes for a role can take valuable time.  Not all applicants will have the required experience or meet the qualifications for the job.  In addition, for reasons of practicality and due to finite bandwidth, there are occasions where the assessment of résumés is curtailed once a sufficient number of [initially] suitable-looking candidates have been identified leaving many résumés on file but in an unreviewed state. Embedded AI can save recruiters and hiring managers time by objectively comparing job requirements with candidate résumés and proactively surfacing candidate recommendations for consideration. These recommendations based on experience, skills and qualifications can reduce the time it takes to find candidates for an open requisition. Learn more about using AI to review high volumes of resumés
  • Engaging candidates and sourcing from within the organization. Keeping candidates engaged and aware of opportunities at your company can expedite finding the right candidate for the right role.  Consider a candidate who could be a great fit for a number of roles, or imagine being able to find the right fit from within your organization. Embedded AI can provide tailored job recommendations that share similar skill, experience and education requirements, helping candidates see beyond one open job. A candidate applying for a job can also be presented with recommendations of additional skills to add to their job application helping them better represent themselves in the hiring process. Through embedded AI, candidates can more easily find and apply to jobs that they are a fit for. Learn more about recommending jobs that match a candidate's profile
  • Find candidates similar to other candidates. Let’s say that a hiring manager has 10 open roles in their organization. The hiring manager who just hired for one of these roles asks their recruiter to find candidates similar to the new hire for the remaining roles. Embedded AI can objectively compare the new hire’s experience and qualification with those of the other candidates, efficiently surfacing qualified candidates for review. Learn more about finding similar candidates

We architected these AI solutions with the goal of providing tangible value to HR teams quickly and efficiently.  This extends to an efficient Implementation that completes in just a few days. Designed as part of the HCM suite, our embedded AI solutions are part of the UX that recruiters, candidates and hiring managers are already familiar with.  By providing meaningful recommendations and predictions on Day 1, customers are quickly on their way to realizing ROI.

Conclusion

Our customers have found that this simplicity in implementation, time to value and ease of use
makes AI a seamless part of the end to end recruiting lifecycle While our AI solutions cannot be setup with the wave of a magic wand (yet), these principles continue to guide new product development and enhancements. 

For more information on Oracle AI, see Oracle Artificial Intelligence.

April Hawthorn

Sr. Director, Product Management, AI Apps for HCM


Previous Post

How to invoke an OCI Model Deployment from OCI APEX Service.

Łukasz Szarabajko | 5 min read

Next Post


Smart Manufacturing- Steel Refractory using OCI Data Lakehouse